Zhou et al. Light: Science & Applications (2018) 7:32 Official journal of the CIOMP 2047-7538DOI 10.1038/s41377-018-0030-0 www.nature.com/lsa

ART ICLE Open Ac ce s s

Single-shot BOTDA based on an opticalchirp chain probe wave for distributedultrafast measurementDengwang Zhou 1, Yongkang Dong 1, Benzhang Wang1, Chao Pang1, Dexin Ba1, Hongying Zhang2, Zhiwei Lu1,Hui Li3 and Xiaoyi Bao 4

AbstractBrillouin optical time-domain analysis (BOTDA) requires frequency mapping of the Brillouin spectrum to obtainenvironmental information (e.g., temperature or strain) over the length of the sensing fiber, with the finite frequency-sweeping time-limiting applications to only static or slowly varying strain or temperature environments. To solve thisproblem, we propose the use of an optical chirp chain probe wave to remove the requirement of frequency sweepingfor the Brillouin spectrum, which enables distributed ultrafast strain measurement with a single pump pulse. Theoptical chirp chain is generated using a frequency-agile technique via a fast-frequency-changing microwave, whichcovers a larger frequency range around the Stokes frequency relative to the pump wave, so that a distributed Brillouingain spectrum along the fiber is realized. Dynamic strain measurements for periodic mechanical vibration, mechanicalshock, and a switch event are demonstrated at sampling rates of 25 kHz, 2.5 MHz and 6.25 MHz, respectively. To thebest of our knowledge, this is the first demonstration of distributed Brillouin strain sensing with a wide-dynamic rangeat a sampling rate of up to the MHz level.

IntroductionIn modern industry1–3, geophysical research4–6, the

health monitoring of civil infrastructures1,2, and themotion capturing of robot hands7,8 and the human body9,a truly distributed ultrafast measurement, are widelyrequired for real-time monitoring of distributed strain ortemperature information. Compared with conventionalelectric sensing networks, optical fiber sensing offers anattractive solution due to several advantages: smallness insize, low cost, long sensing range, chemical inertness, andimmunity to external-electromagnetic interference10,where the optical fiber simultaneously acts as the opticaltransmitting medium and millions of sensing points.

Brillouin-based distributed sensors have been widelyreported since the 1990s2,11,12. Generally, two counter-propagating optical waves (i.e., pump wave and probewave) are launched into a fiber under test (FUT). As thefrequency difference between these two waves approachesthe local Brillouin frequency shift (BFS) of the FUT, theoptical power transferred from the high-frequency lightwave to the low-frequency light wave reaches a maximum.As a result, the Brillouin gain spectrum (BGS) can beobtained by sweeping the detuned frequency. Then, theBFS, which is dependent on both applied strain and tem-perature along the FUT2,10,11,13–15, can be calculated bycurve fitting the BGS. The process of frequency tuning toobtain the BGS is time-consuming. Distributed fast strainmeasurement has been previously realized using severaltechniques: Brillouin optical correlation-domain analysis(BOCDA)3,16–18, Brillouin optical correlation-domainreflectometry (BOCDR)3,19–25, and Brillouin optical time-domain analysis (BOTDA)12,26–32.

© The Author(s) 2018OpenAccessThis article is licensedunder aCreativeCommonsAttribution 4.0 International License,whichpermits use, sharing, adaptation, distribution and reproductionin any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if

changesweremade. The images or other third partymaterial in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to thematerial. Ifmaterial is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtainpermission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Correspondence: Yongkang Dong (aldendong@gmail.com)1National Key Laboratory of Science and Technology on Tunable Laser, HarbinInstitute of Technology, 150001 Harbin, China2Department of Optoelectronic Information, Science and Engineering, HarbinUniversity of Science and Technology, 150080 Harbin, ChinaFull list of author information is available at the end of the article.

1234

5678

90():,;

1234

5678

90():,;

1234567890():,;

1234

5678

90():,;

By employing a BOCDA with optimized time gates andan unbalanced Mach–Zehnder delay line33, a vibrationmeasurement with a frequency of 200Hz and 10-cm spatialresolution in a 20-m measurement range is realized at a 1-kHz sampling rate. The sampling rate can be furtherimproved to 5 kHz based on random accessibility and theuse of a lock-in amplifier34. An improved BOCDR scheme20

is implemented by utilizing a high-speed voltage-controlledoscillator to convert the BFS to a phase delay, resulting in a100-kHz sampling rate at a single point and 100-Hz repe-tition rate at 1000 points with 10 times averaging. In theBOCDA/BOCDR techniques, the high-speed sampling ratefor dynamic strain at a single point is obtained at theexpense of distribution due to the slow sweeping peaklocation process, which requires priori information for thestrain location.Remarkably, BOTDA has been widely studied because of

its superior performance for long-distance-distributedmeasurement27,35–38, wide-strain dynamic range12,31, andhigh-speed sampling rate12,30,31,39–41. In a conventionalBOTDA system, the distributed BGS is reconstructed byinjecting a high-power pump pulse to interrogate the local-stimulated Brillouin scattering (SBS) and then sweeping thecontinuous probe wave frequency over a wide spectralrange; such a process imposes the main limiting factor onthe sampling rate42. To date, BOTDA for dynamic mea-surement is classified into three categories: frequency-comb-based sweep-free scheme43–47, slope-assistedscheme30,31,39,42,48,49, and fast-frequency sweepingscheme50,51. The sweep-free scheme makes use of a fre-quency comb to demodulate the BGS; however, there is atrade-off between the spatial resolution and the frequencyinterval for the comb, with the spatial resolution usually afew tens of meters. In the slope-assisted scheme, the fre-quency detuning between pump and probe waves is set atthe middle of the slopes for the BGS39,40,42, Brillouin phase-shift spectrum30,52 or their ratio spectrum31,49, so that asingle pump pulse can demodulate the distributed BFSalong the fiber. It is noted that the sampling rate is onlylimited by the length of the FUT without averaging, but thedynamic range is restricted by the linewidth of the BGS.Several improved schemes have been introduced toimprove the dynamic range to 5000 με by use of thefrequency-agile technique12,31,53 at the expense of thesampling rate. The last scheme is realized by compressingthe frequency switching time for the probe wave. A 100-Hzvibration measurement in a 100-m FUT based on a fastBOTDA50 is acquired by the frequency-agile technique toquickly switch 100 scanning frequencies for the probe wavecorresponding to 100 synchronous pump pulses, resultingin a sampling rate of ~10 kHz, with the distributed BGStailored from the time trace. Obviously, the sampling ratefor this scheme depends on both the length of the FUT andthe number of scanning frequencies. Recently, a frequency-

sweep pulsed BOTDA51 was implemented to obtain aLorentzian-shaped correlated gain in the time domain bylinear-frequency modulation of two 1.0-μs pulses as pumpand probe pulses coupled with a point (20m) dynamicstrain measurement with a sampling rate of 10 kHzobtained at the end of 10 km. Although a distributeddynamic measurement can be realized through adjustingthe delay between the probe and pump pulses, the effectivesampling rate is divided by the sensing point number.In this paper, we theoretically and experimentally report a

novel single-shot BOTDA for distributed ultrafast mea-surement based on optical chirp chain (OCC), namedOCC–BOTDA, where the probe wave is frequency modu-lated into short optical chirp segments and then cascadedinto an OCC. Only a single-shot pump pulse is required torecover the distributed BGS along the OCC probe wavewhen the local BFS is covered by the wide-frequency spanbetween the pump pulse and the OCC probe wave. Themaximum sampling rate can reach the order of MHz, whichis only restricted by the length of the FUT. The dynamicstrain range can be readily tuned by setting the frequencyspan. We further test the distributed ultrafast performanceof this system: a periodic mechanical vibration is monitoredwith a 25-kHz sampling rate; a mechanical shock mea-surement is designed by rapidly stretching a 2-m fibersection via knocking a rotating mechanism, which is cap-tured at a sampling rate of 2.5MHz; a switch event isdesigned to simulate a fast 20-MHz BFS change, which isrecorded at a sampling rate of 6.25MHz.

Materials and methodsOperation principleA typical BOTDA system is based on a “pump-probe”

scheme2,28,31,54–56 such that the power transfer from theoptical pump pulse to the continuous probe wave occursthrough SBS when their frequency difference is set nearthe BFS of the FUT. The evolution of these two waves(indicated by Ap and As, respectively) and an acousticwave (indicated by Q) is described by the followingcoupled-wave equations2,57

∂Ap

∂z þ nc∂Ap

∂t ¼ iωγe2ncρ0

QAs

∂As∂z � n

c∂As∂t ¼ � iωγe

2ncρ0Q�Ap

ΓB � 2iΩð Þ ∂Q∂t þ Ω2B �Ω2 � iΩΓB

� �Q ¼ ε0γeq

2ApA�s

ð1Þ

where ω≡ 2π⋅νp ≈ 2π⋅νs is the angular frequency of theoptical waves, n is the refractive index of the FUT core, c isthe speed of light in the vacuum, γe is the electrostrictivecoefficient, ρ0 denotes the mean density of the FUT core, ε0is the vacuum permittivity, and ΓB= 1/τp is the Brillouingain linewidth. Ω= 2π(νp− νs), q= kp+ ks. ΩB= 2π⋅fBFS,where fBFS= 2nvpVa/c is the local BFS of the FUT and Va is

Zhou et al. Light: Science & Applications (2018) 7:32 Page 2 of 11

the speed of the moving acoustic wave. Generally, the valueof the BFS is ~11GHz for a single-mode fiber at thecommunication wavelength of 1550 nm2,57.The operation principle for the OCC–BOTDA is illu-

strated in Fig. 1a. The optical probe wave is frequencymodulated fromν1 to νN(N is the frequency number) in afew tens of nanoseconds, resulting in a short optical chirpsegment, which can be regarded as the temporal compres-sion of the traditional sweeping frequency method. Then,the probe wave is composed of several short optical chirpsegments, which are referred to as the optical chirp chain(OCC). It is noted that the OCC has two types of on-linkmodes: one is the sawtooth mode (two adjacent opticalshort-chirp segments are cascaded by a head-to-tail cohe-sion), and the other is the triangular mode (two adjacentoptical short-chirp segments are cascaded by a head-to-heador tail-to-tail cohesion). Similar to typical BOTDA systems,the pump wave is shaped into a pulse pattern, with a widththat is narrower than the duration of the short optical chirpsegment. The OCC probe wave and the pump pulse areinjected into an FUT in the opposite direction. When thefrequency difference between the pump pulse and the OCCprobe wave matches the BFS for the FUT (Fig. 1b), the BGScan be recovered along the short optical chirp segment inthe time domain due to the Brillouin amplification.The OCC–BOTDA system is numerically calculated

based on Eq. (1), with the simulation results corre-sponding to the two chirp modes (magenta line), i.e.,“sawtooth” mode and “triangular” mode shown in Fig. 2a,b, respectively. Here, the duration of the short opticalchirp segment is set at 20 ns, with 300MHz of frequencycoverage, while the width of the pump pulse is set at 10 ns.A 10-m-long fiber with fBFS= 10.70 GHz and ΓB/2π= 30MHz is used. It is clearly illustrated that the intrinsicBGSs (black dashed-dotted line) along the fiber describesa symmetrical Lorentzian profile as the frequency of theOCC is scanned from 10.550 GHz to 10.845 GHz in every2-m segment. Compared with the intrinsic BGS, in boththe “sawtooth” mode and “triangular” mode, the simu-lated BGS exhibits an asymmetrical and broadened pro-file, while its peak shows a time delay of ~2.7 ns (orposition shift) corresponding to a frequency shift of ~41MHz. Such a phenomenon arises from the transient SBS

interaction, i.e., the hysteresis effect due to Brillouinamplification with respect to the acoustic wave excitation.As shown in Fig. 2c, the peaks of the intrinsic BGS and thederivative of the simulated BGS (green line) show com-plete coincidence, which indicates that the growth rate ofthe acoustic wave reaches a maximum for the resonantfrequency and then decreases due to the increaseddetuning from the BFS; subsequently, the measured BGSexhibits a time delay with respect to the intrinsic case.Note that a “ghost peak” appears at the frequency dis-

continuity position in the “sawtooth” mode in Fig. 2a dueto the generation of an equivalent frequency near the BFSwhen the frequency is decreased from the highest fre-quency component νN to the lowest frequency componentν1; the “ghost peak” disappears in the “triangular” mode inFig. 2b due to the smooth convergence for the frequencyin the link point between two short optical chirp seg-ments. In the following experiment, the “sawtooth” modeOCC probe wave is adopted, since the demodulationprocess for the local BFS in the “sawtooth” mode is sim-pler than in the “triangular” mode.

Sampling rate and spatial resolutionIn contrast to the time-consuming process of frequency

sweeping for traditional BOTDA, the distributed BGSalong the FUT based on the OCC–BOTDA can be rapidlydemodulated by only injecting a single-shot pump pulseinto the fiber. Therefore, the maximum sampling rate ofthis scheme is given by

fSa ¼ 1Nave � T ð2Þ

where Nave is the number of averages, T= 2nL/c is theround-trip time for the sensing fiber with a length of L. Itcan be seen that without averaging, the maximum sam-pling rate is only limited by the FUT length.

The spatial resolution for this scheme is restricted bythe duration of the short optical chirp segment ΔT and isgiven by

ΔzSR ¼ cΔT2n

ð3Þ

Pumpa

BGSFUT

OCC

ChirpProbe

Probe Pump

SBS

ƒBFS

�N�1�2

�N

L

�1�2

�p�

b

Fig. 1 The operation principle of an OCC-BTODA. a The OCC probe wave is temporally cascaded by several short optical chirp segments in whichthe frequency is scanned from v1 to vN, while the pump pulse requires just single shot to obtain the distributed BGS along the FUT. b The frequencyrelationship between the pump pulse and the continuous probe wave

Zhou et al. Light: Science & Applications (2018) 7:32 Page 3 of 11

0

1.2

a

Inte

nsity

(a.

u.)

Inte

nsity

(a.

u.)

1.0

0.8

0.6

0.4

0.2

1.2

1.0

0.8

0.6

0.4

0.2

0.0

1.2 BGS intrinsicBGS simulatedDerivative of the BGS simulated

1.0

0.8

Inte

nsity

(a.

u.)

0.6

0.4

0.2

0.00.0 0.5 1.0

Position (m)

1.5 2.0

0 2 4 6 8 10Position (m)

0.00 2 4 6

Position (m)8

ChirpBGS simulatedBGS intrinsic

ChirpBGS simulatedBGS intrinsic

10

20 40 60Time (ns)

Main peakGhost peak

80 100

10.90

10.90

10.85

10.80

Freq

uenc

y (G

Hz)

Inte

nsity

(a.

u.)

10.75

10.70

10.65

10.60

10.55

3

2

1

0

–1

–2

–3

10.85

10.80

Freq

uenc

y (G

Hz)

10.75

10.70

10.65

10.60

10.55

0 20 40

Main peak

60Time (ns)

80 100

0 4 8 12Time (ns)

16 20

b

c

Fig. 2 Simulation of the OCC–BOTDA. a The distributed BGSs along the fiber for the intrinsic case (black dashed-dotted line), “sawtooth” mode(blue line), and distributed chirp frequency (magenta line). b The distributed BGSs along the fiber for the intrinsic case (black dashed-dotted line),“triangular” mode (blue line), and distributed chirp frequency (magenta line). c The intrinsic BGS (black dashed-dotted line), simulated BGS (blue line),and derivative of the simulated BGS (green line) within the first 2 m

Zhou et al. Light: Science & Applications (2018) 7:32 Page 4 of 11

Experimental setupFigure 3 shows the experimental setup used to test the

ultrafast performance of the OCC–BOTDA scheme. Thefrequency-agile technique involves generation of fast-frequency-changing microwaves by an AWG with a highbandwidth and high digital sampling rate, in which theoutput can be reconfigured by editing the AWG controlprogram. The frequencies should satisfy the equation53

fAWG tð Þ ¼ f1 þ η � t ð4Þ

where f1 is the initial frequency. η=Δf/ΔT is thefrequency-changing rate, Δf is the frequency span, and ΔTis the duration of the short optical chirp segment. Due tothe bandwidth limitation for AWG in our laboratory, wechose the dual modulation53 to implement a wide-frequency span between the OCC probe wave and thepump pulse, which can cover the BFS of the FUT.An optical fiber laser with a narrow linewidth of ~10 kHz

was used as the light source, in which the output parameterswere set at 1550 nm and 100mW. Subsequently, the 90%branch (red line) for the optical coupler was selected to bechopped for the optical pulse as the pump wave. The lightwas frequency modulated using an electro-optic modulator1in the carrier-suppressed regime, which is driven by an 8.15-GHz sinusoidal microwave output from a microwave gen-erator. Then, the frequency-modulated light is intensitymodulated using EOM2 driven by the CH2 of the AWGinto a 10-ns optical pulse. After amplification to 1W by anerbium-doped fiber amplifier1, an optical pulse with ahigher-frequency sideband was selected through an opticalfiber Bragg grating (FBG) filter. Then, the optical pulse wasinjected into the FUT through an optical circulator.The lower branch with 10% component was used to gen-

erate an OCC for use as a probe wave. The light was fre-quency downshifted using EOM3 driven by the CH1 ofAWG and a tunable FBG. Note that the output of CH1 wastemporally cascaded by the sinusoidal microwave signals

with a frequency span from 2.4 to 2.695GHz and a frequencystep of 5MHz. As a result, a short optical chirp segment wasgenerated with a width of 20 ns and frequency-changing rateof η= 15MHz/ns. By using the “sawtooth” mode, the OCCprobe wave was assembled and then amplified by EDFA2 to~80 μW. Hence, both the pump pulse and the continuousprobe wave were launched into the FUT. Finally, the Bril-louin signal (i.e., amplified OCC-probe wave) was received bya D&A module at a sampling rate of 5GS/s. To eliminatepolarization noise and fading, the polarization of all opticaldevices was aligned along the slow axis, with a polarizationmaintaining fiber (PMF) with a BFS of 10.705GHz used asthe FUT. A photodetector with a bandwidth of 350MHz anda response time of 2 ns in the D&A module was used todetect the OCC-probe wave.

Results and discussionsStatic BGS measurement and twice-correlation algorithmThe measured time traces of the Brillouin signal of the

OCC–BOTDA system are shown in Fig. 4a. The dataclearly show a BGS with a main peak and a ghost peak thatis found to occur every 20 ns (corresponding to a spatialresolution of 2m), which agrees well with the simulationresults, as plotted in Fig. 2. Compared with the BGS (blackline) for a strain change of Δε= 0.0 με, the BGS (magentaline within the blue wireframe) for a strain change Δε= 700με is time-shifted. Subsequently, a zoom-in view for theblue wireframe is shown in Fig. 4b as the strain changesfrom Δε= 0.0 με to Δε= 700 με, with the horizontal axislabel converted from a time value to a frequency value. It isclearly shown that the central frequency of the main peakinitially shifts by ~42MHz with respect to the local BFS at~10.705 GHz at room temperature. The time (or frequency)of the main peak is evidently up-shifted as the strain changeis increased.However, it is very difficult to precisely calculate the

BFS by typical Lorentzian or Gaussian fitting the main

Fig. 3 Experimental configuration for the OCC-BTODA system. MG microwave generator, EOM electro-optic modulator, AWG arbitrary-waveformgenerator, EDFA erbium-doped fiber amplifier, FUT fiber under test, TFBG tunable fiber Bragg grating, OC optical coupler, OI optical isolator, OCiroptical circulator, D&A photodetector and data acquisition modules, FB fixed base. Note that the polarization for all the optical devices is alignedalong the slow axis to eliminate polarization noise and fading

Zhou et al. Light: Science & Applications (2018) 7:32 Page 5 of 11

peak of the BGS due to its asymmetrical profile. Here, wepropose a twice-correlation algorithm not only to makethe measured BGS more symmetrical, but also to improvethe signal-to-noise ratio (SNR)58,59. Based on the intrinsic

mathematical property of the correlation function, thisalgorithm can be used to realize two functions. One is thesymmetric correlation peak generated by the auto- orcross-correlation between the BGSs, which enables a

1.2

a

Bril

loui

n si

gnal

s (a

.u.)

0.8

0.4

0.0

240 280 320 360 400 440 480 520 560 600 640

Δ�= 0.0 με Δ�= 700 μεΔ�= 0.0 μεΔ�= 233 μεΔ�= 467 μεΔ�= 700 με

Δ�= 584 μεΔ�= 350 μεΔ�= 116 με

680 720

1.5

1.0

Inte

nsity

(a.

u.)

0.5

0.010.55 10.60 10.65 10.70 10.75 10.80 10.85

Frequency (GHz)Time (ns)

b

Fig. 4 The static distributed BGS measurement based on the proposed OCC-BOTDA. a The time traces for the Brillouin signals for a strainchange of Δε= 0.0 με and Δε= 700 με. b The zoom-in view for the BGSs for a strain change from Δε= 0.0 με to Δε= 700 με in the blue box.Averaging is carried out 200 times

1.0b

c

d

0.8START

END

Input BGS0 and BGS

Reference?Yes No

BGS1=BGS BGS0

BGS2=BGS1 BGS01

BGS02=BGS01 BGS01

ƒBFS0=Lorentzfitting (BGS02)

ΔƒBFS=ƒBFS–ƒBFS0

Output ΔƒBFS

ƒBFS=Lorentzfitting (BGS2)

BGS01=BGS0 BGS0

Inte

nsity

(a.

u.)

ΔƒB

FS (

MH

z)Δƒ

BF

S (

MH

z)

0.6

0.4

0.2

403530252015105

–50 10 20

Linear fitting

30 40 50

0

40

35

30

25

20

15

10

5100 200

0.0–600 –400 –200 0 200 400 600

Frequency (MHz)

Position (m)

300 400 500Δ� (με)

600 700

ΔƒBFS

Δ�= 0.0 με

Δ�= 116 με

Δ�= 233 με

Δ�= 350 με

Δ�= 467 με

Δ�= 584 με

Δ�= 700 με

Δ�= 700 με

a

Fig. 5 The strain change demodulation process. a The demodulation flow chart based on a twice-correlation algorithm; ⋆ represents the cross-correlation operator. b The reference BGS and the random BGS are processed by this algorithm, resulting in two symmetrical profiles. c Thedemodulated BFS changes along the 50-m PMF. d Change in BFS versus change in strain

Zhou et al. Light: Science & Applications (2018) 7:32 Page 6 of 11

much easier precise calculation for the main peak loca-tion. The other function is the position or frequency shiftof the correlation peaks for different strains, corre-sponding to changes in BFS or strain.The demodulation flow chart for this algorithm is

shown in Fig. 5a. First, a measured BGS in the looseregion is selected as the reference BGS0, and then, its firstautocorrelation peak BGS01 and second autocorrelationpeak BGS02 (black line in Fig. 5b) are computed. Thetwice-correlation results for the measured BGSs are givenby

BGS2 ¼ BGS ? BGS0ð Þ ? BGS01 ð5Þ

where ⋆ represents the cross-correlation operator. It canbe seen that both correlation peaks BGS02 and BGS2 (e.g.,magenta line for Δε= 700 με) show more symmetricalprofiles and higher SNR compared to the original BGSs inFig. 4b. Next, the correlation peaks BGS02 and BGS2 areLorentzian or Gaussian fitted to calculate their centralfrequency fBFS0 and fBFS. Finally, the output of this flowchart is the BFS change

ΔfBFS ¼ fBFS � fBFS0 ð6ÞAs a result, the distributed BFS changes for different

strain measurements are shown in Fig. 5c; the data showthat the demodulated BFS change in the strain segment at10–12 m is clearly migrated away from a value of

approximately zero in the non-strain segment. The straindependences for the demodulated BFS changes are plot-ted in Fig. 5d, which reveals a good linearity with a fittingstrain coefficient of 0.049MHz/με.

Dynamic strain measurementsTo test the dynamic measurement performance, three

different types of dynamic strain are established asfollows:

Periodic mechanical vibration measurementA 2-m segment located at a position of 44–46m for a

50-m FUT is driven by an electro-motor to induce aperiodic mechanical vibration for the test. The frequency-changing rate η is set to 20MHz/ns, while the durationfor the short optical chirp segment is set to 20 ns. For dataacquisition, the original measured time trace for theBrillouin signal with no averaging can be collected via thefast-frame mode while setting the sampling rate to 25 kHz(corresponding to a period of 40 μs).According to the order of frame and time, the time trace

is reshaped with frame length Nseg= 3200(>2nL/c⋅5GS/s),forming a 3200 × 11000 array. The time evolution for thedistributed BGS is shown in Fig. 6a. The segment of60–62 m without signal, as shown in Fig. 6b, correspondsto a region where no sensing fiber is present. As shown inFig. 6c, the BGSs at the vibrated segment of 44–46m are

60

60

40

Pos

ition

(m

)

30

20

10

0 4

5

6

60

60

50

40

30

20

10

0200 400

61

62

44

45

46

200

Time (ms) Time (ms) Time (ms)400 50 100 150 200 250 300 350 400

50 100 150 200 250 300 350 400

50 100 150 200 250

ba

300 350 400

c

d

e

Fig. 6 The position-time response for the periodic mechanical vibration. a The time evolution for the distributed BGS (the value is increasedfrom blue to red). b The background pattern at the segment of 60–62 m. c The time evolution for the BGS at the segment of 44–46 m. d The timeevolution for the BGS at the segment of 4–6 m. e The time evolutions for the distributed BGSs are processed by subtracting the initial distributedBGSs at time t= 0

Zhou et al. Light: Science & Applications (2018) 7:32 Page 7 of 11

position-shifted over time, while the BGSs at the loosesegment of 4–6m remain unchanged in Fig. 6d. Subse-quently, by subtracting the initial distributed BGSs at timet= 0, the vibration position span and periodic patterns arerapidly processed, which can then be clearly located asdemonstrated in Fig. 6e, which is suitable for real-timeidentification of vibration locations. The green line inFig. 7a shows the time evolution for the strain change atthe segment of 44–46m demodulated by the “twice-cor-relation algorithm,” with the red line showing the result ofa 30-points moving average, which corresponds to thewaveform of the periodic mechanical vibration. The fre-quency spectrum for this waveform is obtained throughfast Fourier transform, as shown in Fig. 7b, which containsa fundamental frequency component of 31.8 Hz withharmonics of 62.0 and 93.2 Hz.

Mechanical shock measurementA mechanical shock measurement is designed, as shown

by the schematic diagram in Fig. 8. A 2-m segment of a10-m FUT is selected to be stretched. The two end pointsfor this section are fixed onto FB1 and the free end of a

lever. In this experiment, through rapidly impacting thelever by a hammer, the free end of the lever will be rotatedaround FB2 so that the 2-m fiber section is quicklystretched. Meanwhile, this mechanical shock is capturedby the proposed OCC–BOTDA system in which thesampling rate is set at 2.5MHz (corresponding to a periodof 400 ns). The frequency-changing rate η is 15MHz/ns,while the duration for the short optical chirp segment isset to 20 ns.The time evolutions for the BGSs in this section are first

processed by the “twice-correlation algorithm,” with theresult plotted in Fig. 9a. The demodulated strain changeinduced by mechanical shock is shown in Fig. 9b, wherethe green line shows the original result with no averagingand the red line shows the result obtained for a 30-pointsmoving average. A shock time of ~250 μs with a strainchange of ~800 με is demonstrated from the no-strainstate at time t= ~400 μs to the strain state at timet= ~750 μs. This result shows tremendous potential forcapturing fast processes, e.g., an explosion process. Thestandard deviation for the result is ~120 με with noaveraging, while it decreases to 35 με for 30-points mov-ing averaging since the power magnification of EDFA2decreases as the repetition rate of the pump pulseincreases to the order of a MHz.

Switch event demoA switch event is carried out to demonstrate the

ultrafast acquisition capability of the OCC–BOTDA sys-tem. The frequency-changing rate η and the duration forthe short optical chirp segment are set to 15MHz/ns and20 ns, respectively, as done previously. Its notion can besummarized by noting that the microwave frequencyagility spans every adjacent optical chirp segment are setat 2.400–2.695 GHz and 2.380–2.675 GHz, resulting in afrequency shift of 20MHz (i.e., a strain change of ~408

800No averaging30 pts moving average

31.8 Hz

DC

62.0 Hz

93.2 Hz

600

400

Δ� (μ

ε)

Inte

nsity

(a.

u.)

200

0

–200 0.0

0.2

0.4

0.6

0.8

1.0

ba

0 160 240 320 400 0 25 50 75 100 125 150 175 200Time (ms) Frequency (Hz)

80

Fig. 7 The time and frequency distribution of the periodic mechanical vibration. a The time evolution for the strain changes (green line) withno averaging, which represents the waveform for the periodic mechanical vibration, and 30-points moving average result (red line). b The frequencyspectrum of the vibration

Stop metope

Lever

30 cm

0 m1 m

2 m

< - Probe

Pump - > 20 cm

10 cm

0 cm

FUT

Hammer

FB2

FB1

Fig. 8 Schematic diagram for the mechanical shock measurement

Zhou et al. Light: Science & Applications (2018) 7:32 Page 8 of 11

με), which can be used to simulate a quick BFS changeover time. The repetition rate for the pump pulse canreach up to 6.25MHz (corresponding to a period of 160ns).The time evolution for the BGSs is processed by the

“twice-correlation algorithm” at 200 times averaging andno averaging, as shown in Fig. 10a, b, respectively. TheirBFS changes over time are shown in Fig. 10c, where thelower and the upper scatter sequences correspond to thefrequency-agility spans of 2.400–2.695 GHz and2.380–2.675 GHz, respectively. The frequency differencebetween the two scatter sequences is ~20MHz, with theerror for averaging 200 times and no averaging lyingwithin ±0.5MHz and ±2.5MHz, respectively.

ConclusionsIn this work, an OCC–BOTDA scheme was theoreti-

cally proposed, numerically simulated, and experimentallyverified for distributed ultrafast measurement with a

maximum sampling rate of 6.25MHz. The OCC probewave is composed of several short optical chirp segments,which are generated through frequency modulation byuse of the frequency-agile technique. When the frequencyspan between the OCC probe wave and the pump pulseoverlaps the BFS for the sensing fiber, the distributed BGSalong the fiber is revealed via the OCC probe wave in thetime (or position) domain using only a single-shot pumppulse. The sampling rate for this proposed OCC–BOTDAis only restricted by the fiber length.The performance for this proposed OCC–BOTDA is

limited by the trade-offs among the frequency span, spatialresolution, and SNR. The effective scanning range is lessthan the frequency span due to the influence of a “ghostpeak” and broadened BGS profile. As the frequency spanincreases for a wide-dynamic range, the SBS interactionregion will be compressed, resulting in recovery of the BGSover a narrow region with a deteriorating SNR. The rela-tively lower SNR in this work may be due to the following

200a 1200No averaging30 pts moving average

800

400

0

–400

100

0

–100

–2000 200 400 600

Time (μs)

800 1000 200

No strain

Shock

Strain

~250.0 μs

400 600 800 1000

Time (μs)

0

Δƒ (

MH

z)

Δ� (μ

ε)

b

Fig. 9 Time–frequency response for the mechanical shock measurement. a The BGS over time processed by the “twice-correlation algorithm”and b its strain change over time with no averaging (green line) and the result obtained for a 30-points moving average (red line)

a c

b

100 30

25

20

15

10

5

0

–50.0 0.5 1.0 1.5

Time (μs)

2.0 2.5 3.0 3.5

0

–100

–100

100

0

0 1 2

Time (μs)

3

1 2 3

0

Δƒ (

MH

z)

ΔƒB

FS (

MH

z)

200 times averagingNo averaging

Fig. 10 The time–frequency response for the switch event. The BGS over time is processed by the “twice-correlation algorithm” for a averaging200 times and b no averaging. c The BFS changes over time for averaging 200 times (black scatter) and no averaging (red scatter)

Zhou et al. Light: Science & Applications (2018) 7:32 Page 9 of 11

factor: the AWG used in this work is composed by syncinga single output channel AWG with a high-sampling rate toan arbitrary-function generator with a lower-sampling rate,which results in electrical signal jitter with a root-mean-square of ~100 ps between the pump pulse and the OCCprobe wave. For future work, a high-performance dual-channel AWG can be used to generate the pump pulse andthe OCC probe wave to increase the SNR.Additionally, the D&A module can be followed by a

field-programmable gate array (FPGA) to realize a real-time response. We believe that the proposedOCC–BOTDA can provide dynamic or real-time dis-tributed measurement for many potential applications,e.g., monitoring of large infrastructure and capturing anexplosion process.

AcknowledgementsThis work was supported by the National Key Scientific Instrument andEquipment Development Project of China (2017YFF0108700) and NationalNatural Science Foundation of China (61575052). We like to express ourgratitude to P.B. Xu and T.F. jiang for their suggestion in the primaryexperiment and P.L. Tong and L.W. Sheng for their contribution to thediagrams for the experimental setup.

Author details1National Key Laboratory of Science and Technology on Tunable Laser, HarbinInstitute of Technology, 150001 Harbin, China. 2Department of OptoelectronicInformation, Science and Engineering, Harbin University of Science andTechnology, 150080 Harbin, China. 3School of Civil Engineering, HarbinInstitute of Technology, 150001 Harbin, China. 4Fiber Optics Group,Department of Physics, University of Ottawa, Ottawa, ON K1N 6N5, Canada

Conflict of interestThe authors declare no conflict of interest.

Received: 13 September 2017 Revised: 13 April 2018 Accepted: 7 May 2018Accepted article preview online: 11 May 2018

References1. Bao, X. Y. & Chen, L. Recent progress in distributed fiber optic sensors. Sensors

12, 8601–8639 (2012).2. Bao, X. Y. & Chen, L. Recent progress in brillouin scattering based fiber sensors.

Sensors 11, 4152–4187 (2011).3. Hotate K. Recent achievements in BOCDA/BOCDR. Proceedings of IEEE SEN-

SORS 142–145 (IEEE, Valencia, Spain, 2014).4. He, Z. Y., Liu, Q. W. & Tokunaga, T. Ultrahigh resolution fiber-optic quasi-static

strain sensors for geophysical research. Photon. Sens. 3, 295–303(2013).

5. Mirhabibi, M., Negarestani, A., Bolorizadeh, M. A., Rezaei, M. R., & Akhound, A. Anew approach for radon monitoring in soil as an earthquake precursor usingoptical fiber. J. Radioanal. Nucl. Chem. 301, 207–211 (2014).

6. Zhang, W. T., Huang, W. Z., Li, L., Liu, W. Y. & Li, F. High resolution strain sensorfor earthquake precursor observation and earthquake monitoring. Proceedingsof the 6th European Workshop on Optical Fibre Sensors (SPIE, Limerick, Ireland,2016).

7. Leo, J., Low, K., Costa, J., Black, R. J. & Park, Y. L. Fiber optically sensorizedmulti-fingered robotic hand. Proceedings of 2015 IEEE/RSJ International Con-ference on Intelligent Robots and Systems 1763–1768 (IEEE, Hamburg, Germany,2015).

8. Xiao L. F. et al. Impedance control of a robot needle with a fiber optic forcesensor. Proceedings of the 13th International Conference on Signal Processing1379–1383 (IEEE, Chengdu, China, 2016).

9. Guo, J. J., Niu, M. X. & Yang, C. X. Highly flexible and stretchable optical strainsensing for human motion detection. Optica 4, 1285–1288 (2017).

10. Zadok, A. et al. Random-access distributed fiber sensing. Laser Photonics Rev. 6,L1–L5 (2012).

11. Kurashima, T., Horiguchi, T. & Tateda, M. Distributed-temperature sensing usingstimulated Brillouin scattering in optical silica fibers. Opt. Lett. 15, 1038–1040(1990).

12. Ba, D. X. et al. Distributed measurement of dynamic strain based on multi-slope assisted fast BOTDA. Opt. Express 24, 9781–9793 (2016).

13. Kurashima, T., Horiguchi, T. & Tateda, M. Thermal effects of Brillouin gainspectra in single-mode fibers. IEEE Photonics Technol. Lett. 2, 718–720 (1990).

14. Niklès, M., Thévenaz, L. & Robert, P. A. Simple distributed fiber sensor based onBrillouin gain spectrum analysis. Opt. Lett. 21, 758–760 (1996).

15. Bao, X., Webb, D. J. & Jackson, D. A. 22-km distributed temperature sensorusing Brillouin gain in an optical fiber. Opt. Lett. 18, 552–554 (1993).

16. Hotate, K. Fiber distributed brillouin sensing with optical correlation-domaintechniques. Proceedings of Asia Communications and Photonics Conference(OSA, Shanghai, China, 2014).

17. Hotate, K., Arai, H. & Song, K. Y. Range-enlargement of simplified brillouinoptical correlation domain analysis based on a temporal gating scheme. SICEJ. Control Meas. Syst. Integr. 1, 271–274 (2008).

18. Song, K. Y., He, Z. Y. & Hotate, K. Distributed strain measurement withmillimeter-order spatial resolution based on Brillouin optical correlationdomain analysis. Opt. Lett. 31, 2526–2528 (2006).

19. Mizuno, Y., Zou, W. W., He, Z. Y. & Hotate, K. Proposal of Brillouin opticalcorrelation-domain reflectometry (BOCDR). Opt. Express 16, 12148–12153(2008).

20. Mizuno, Y., Hayashi, N., Fukuda, H., Song, K. Y. & Nakamura, K. Ultrahigh-speeddistributed Brillouin reflectometry. Light Sci. Appl. 5, e16184 (2016).

21. Lee, H., Hayashi, N., Mizuno, Y. & Nakamura, K. Slope-assisted Brillouin opticalcorrelation-domain reflectometry: proof of concept. IEEE Photonics J. 8,6802807 (2016).

22. Mizuno, Y., He, Z. Y. & Hotate, K. One-end-access high-speed distributed strainmeasurement with 13-mm spatial resolution based on Brillouin opticalcorrelation-domain reflectometry. IEEE Photonics Technol. Lett. 21, 474–476(2009).

23. Lee, H., Hayashi, N., Mizuno, Y. & Nakamura, K. Operation of slope-assistedBrillouin optical correlation-domain reflectometry: comparison of systemoutput with actual frequency shift distribution. Opt. Express 24, 29190–29197(2016).

24. Manotham, S., Kishi, M., He, Z. Y. & Hotate K. 1-cm spatial resolution withlarge dynamic range in strain distributed sensing by Brillouin opticalcorrelation domain reflectometry based on intensity modulation. ProceedingsVolume 8351, Third Asia Pacific Optical Sensors Conference (SPIE, Sydney, Aus-tralia, 2012).

25. Denisov, A., Soto, M. A. & Thévenaz, L. Going beyond 1000000 resolved pointsin a Brillouin distributed fiber sensor: theoretical analysis and experimentaldemonstration. Light Sci. Appl. 5, e16074 (2016).

26. Li, W. H., Bao, X. Y., Li, Y. & Chen, L. Differential pulse-width pair BOTDA for highspatial resolution sensing. Opt. Express 16, 21616–21625 (2008).

27. Soto, M. A., Bolognini, G. & Di Pasquale, F. Long-range simplex-coded BOTDAsensor over 120km distance employing optical preamplification. Opt. Lett. 36,232–234 (2011).

28. Dong, Y. K. et al. High-spatial-resolution fast BOTDA for dynamic strain mea-surement based on differential double-pulse and second-order sideband ofmodulation. IEEE Photonics J. 5, 2600407–2600407 (2013).

29. Urricelqui, J., Sagues, M. & Loayssa, A. Synthesis of Brillouin frequency shiftprofiles to compensate non-local effects and Brillouin induced noise inBOTDA sensors. Opt. Express 22, 18195–18202 (2014).

30. Tu, X. B., Luo, H., Sun, Q., Hu, X. Y. & Meng, Z. Performance analysis of slope-assisted dynamic BOTDA based on Brillouin gain or phase-shift in opticalfibers. J. Opt. 17, 105503 (2015).

31. Zhou, D. W. et al. Slope-assisted BOTDA based on vector SBS and frequency-agile technique for wide-strain-range dynamic measurements. Opt. Express 25,1889–1902 (2017).

32. Soto, M. A., Ramírez, J. A. & Thévenaz, L. Intensifying the response of dis-tributed optical fibre sensors using 2D and 3D image restoration. Nat. Com-mun. 7, 10870 (2016).

33. Song, K. Y. & Hotate, K. Distributed fiber strain sensor with 1-kHz sampling ratebased on Brillouin optical correlation domain analysis. IEEE Photonics Technol.Lett. 19, 1928–1930 (2007).

Zhou et al. Light: Science & Applications (2018) 7:32 Page 10 of 11

34. Zhang, C. Y., Kishi, M. & Hotate, K. 5000 points/s high-speed random acces-sibility for dynamic strain measurement at arbitrary multiple points along afiber by Brillouin optical correlation domain analysis. Appl. Phys. Express 8,042501 (2015).

35. Sun, Q., Tu, X. B., Sun, S. L. & Meng, Z. Long-range BOTDA sensor over 50 kmdistance employing pre-pumped Simplex coding. J. Opt. 18, 055501 (2016).

36. Dong, Y. K., Chen, L. & Bao, X. Y. Time-division multiplexing-based BOTDA over100km sensing length. Opt. Lett. 36, 277–279 (2011).

37. Gyger, F., Rochat, E., Chin, S., Niklès, M. & Thévenaz, L. Extending the sensingrange of Brillouin optical time-domain analysis up to 325 km combining fouroptical repeaters. Proceedings of the 23rd International Conference on OpticalFibre Sensors p91576Q (SPIE, Santander, Spain, 2014).

38. Dong, Y. K., Chen, L. & Bao, X. Y. Extending the sensing range of Brillouinoptical time-domain analysis combining frequency-division multiplexing andin-line EDFAs. J. Light Technol. 30, 1161–1167 (2012).

39. Peled, Y., Motil, A., Yaron, L. & Tur, M. Slope-assisted fast distributed sensing inoptical fibers with arbitrary Brillouin profile. Opt. Express 19, 19845–19854 (2011).

40. Motil, A., Danon, O., Peled, Y. & Tur, M. Pump-power-independent doubleslope-assisted distributed and fast Brillouin fiber-optic sensor. IEEE PhotonicsTechnol. Lett. 26, 797–800 (2014).

41. Minardo, A., Coscetta, A., Bernini, R. & Zeni, L. Heterodyne slope-assisted Bril-louin optical time-domain analysis for dynamic strain measurements. J. Opt.18, 025606 (2016).

42. Bernini, R., Minardo, A. & Zeni, L. Dynamic strain measurement in optical fibersby stimulated Brillouin scattering. Opt. Lett. 34, 2613–2615 (2009).

43. Voskoboinik, A., Yilmaz, O. F., Willner, A. W. & Tur, M. Sweep-free distributedBrillouin time-domain analyzer (SF-BOTDA). Opt. Express 19, B842–B847 (2011).

44. Voskoboinik, A. et al. Frequency-domain analysis of dynamically applied strainusing sweep-free Brillouin time-domain analyzer and sloped-assisted FBGsensing. Opt. Express 20, B581–F586 (2012).

45. Feng, Z. H. et al. Performance-enhanced direct detection optical OFDMtransmission with CAZAC equalization. IEEE Photonics Technol. Lett. 27,1507–1510 (2015).

46. Zhao, C. et al. BOTDA using channel estimation with direct-detection opticalOFDM technique. Opt. Express 25, 12698–12709 (2017).

47. Fang, J., Xu, P. B., Dong, Y. K. & Shieh, W. Single-shot distributed Brillouin opticaltime domain analyzer. Opt. Express 25, 15188–15198 (2017).

48. Urricelqui, J., Sagues, M. & Loayssa, A. BOTDA measurements tolerant to non-local effects by using a phase-modulated probe wave and RF demodulation.Opt. Express 21, 17186–17194 (2013).

49. Yang, G. Y., Fan, X. Y. & He, Z. Y. Strain dynamic range enlargement of slope-assisted BOTDA by using Brillouin phase-gain ratio. J. Light. Technol. 35,4451–4458 (2017).

50. Peled, Y., Motil, A. & Tur, M. Fast Brillouin optical time domain analysis fordynamic sensing. Opt. Express 20, 8584–8591 (2012).

51. Kito, C., Takahashi, H., Toge, K. & Manabe, T. Dynamic strain measurement of10-km fiber with frequency-swept pulsed BOTDA. J. Light Technol. 35,1738–1743 (2017).

52. Urricelqui, J., Zornoza, A., Sagues, M. & Loayssa, A. Dynamic BOTDA mea-surements based on Brillouin phase-shift and RF demodulation. Opt. Express20, 26942–26949 (2012).

53. Ba, D. X. et al. Dynamic distributed Brillouin optical fiber sensing based ondual-modulation by combining single frequency modulation and frequency-agility modulation. IEEE Photonics J. 9, 7102908 (2017).

54. Dong, Y. K. et al. Characterization of evolution of mode coupling in a graded-index polymer optical fiber by using Brillouin optical time-domain analysis.Opt. Express 22, 26510–26516 (2014).

55. Dong, Y. K., Zhang, H. Y., Chen, L. & Bao, X. Y. 2cm spatial-resolution and 2 kmrange Brillouin optical fiber sensor using a transient differential pulse pair. Appl.Opt. 51, 1229–1235 (2012).

56. Xu, P. B. et al. Bend-insensitive distributed sensing in singlemode-multimode-singlemode optical fiber structure by using Brillouin optical time-domainanalysis. Opt. Express 23, 22714–22722 (2015).

57. Boyd, R. W. Nonlinear Optics 429–471 (Academic Press, Burlington, 2008).58. Farahani, M. A., Castillo-Guerra, E. & Colpitts, B. G. Accurate estimation of

Brillouin frequency shift in Brillouin optical time domain analysis sensors usingcross correlation. Opt. Lett. 36, 4275–4277 (2011).

59. Wang, F., Zhan, W. W., Zhang, X. P. & Lu, Y. G. Improvement of spatial reso-lution for BOTDR by iterative subdivision method. J. Light Technol. 31,3663–3667 (2013).

Zhou et al. Light: Science & Applications (2018) 7:32 Page 11 of 11